Positioning and release date
A version of the Google Cloud Document AI Enterprise Document OCR processor, used to recognize and extract text and layout information from documents. Google officially announced its release as a Release Candidate on February 3, 2025; the current processor list identifies it as public preview and states that it is similar to v2.1. (docs.cloud.google.com)
Capabilities and uses
- Recognizes printed and handwritten text in documents and extracts text and layout information; the processor supports more than 200 languages. It can be used for digitizing scanned documents, document search, and subsequent data processing. (docs.cloud.google.com)
- Supports configurations including native PDF text extraction, language detection, image quality scoring, and processing of specified pages. This version also supports optional OCR add-ons: LaTeX extraction of mathematical formulas, checkbox state recognition, and font style detection. (docs.cloud.google.com)
- Common supported formats include PDF, GIF, TIFF, JPEG, PNG, BMP, and WebP. (docs.cloud.google.com)
Limitations
- This version is a Release Candidate in public preview, not a stable version marked as generally available (GA) in the processor list. The official version list states that it is not available in
US,EU, andasia-southeast1; the 2025 announcement initially listed the available regions asasia-south1,australia-southeast1,europe-west2,europe-west3, andnorthamerica-northeast1. (docs.cloud.google.com) - The per-document limits listed in the processor documentation are 15 pages for synchronous requests and 500 pages for asynchronous batch processing; synchronous requests can handle up to 30 pages when imageless mode is enabled. Specified page ranges are supported only for synchronous processing. (docs.cloud.google.com)
- Image quality analysis may produce false positives for defect-free digital documents and is better suited to scanned or photographed documents; enabling it increases processing latency. (docs.cloud.google.com)
Pricing
Date checked: November 19, 2026. The official published prices are in US dollars (USD) for the Enterprise Document OCR processor, billed by pages processed, rather than separate token prices for this version: in the standard pricing column, the first 1,000 pages per month are free; the 1,000 to 5,000,000 page tier is USD 1.50/1,000 pages; and the 5,000,000 pages and above tier is USD 0.60/1,000 pages. The standard price for optional OCR add-ons is USD 6.00/1,000 pages. Input and output are not billed separately by token; actual charges and regional applicability should be confirmed against the official pricing page and bill. (cloud.google.com)